Thought I'd share something one of my behaviourism lecturers told me, before we started our semester long flash card SRS project*:

>

You are being trained to write characters, not to make whatever facial expression/noise you make when "thinking / trying to remember something". It is quite clear if you don't remember it - you will know within 1-2 seconds of being prompted. If the desired response does not come immediately, reveal the answer, say it aloud (important) and move on.

Key:

If you don't know it, press reveal straight away, say it aloud, move on to the next one.

Whether you may have *got it* after a short period of thinking is doesn't matter: the goal is to instantly recognize / be able to write / define a word.

Do this and you will avoid training irrelevant behaviours, and focus on improving your ability to instantly recall the desired response.

*We had to memorize 60+ behaviourism terms to the point where we could recall at least 30 with less than 5% error in one minute. We used flash cards and recorded our progress, quite similar to Skritter's method (only without the auto-SRS)

There are multiple levels of knowing something.

1. You don't know it at all.
2. When someone asks the question, you don't know the answer, but you recognize the answer when you hear it.
3. When someone asks the question, you can produce the answer.

These are just points on a continuum and your memory can be somewhere in between.

One question is whether you will learn better if you allow yourself to spend 10 or 20 seconds thinking about the answer, or if you give yourself a short cutoff time like your lecturer did. This is an empirical question and could be answered by doing a small experiment: Try to memorize 100 new items using a short cutoff time, and 100 new items with a longer cutoff time. Study each group for a fixed amount of time. Then test yourself a day later and see how many you remember from each group.

I could go either way on this question. On the one hand I often find that thinking for a few seconds more allows me to retrieve memories that are somewhere between level 2 and level 3, and the next time I am prompted with that item, the memory tends to be solid. On the other hand, using a short cutoff time allows you to review more items per minute.

James

Thats an interesting question, I might experiment and see how it affects my weekly stats (:

I partially agree with faceleg

I don't spend much time busting my brain for a character I don't remember, but I don't full on hit the reveal.

I like to hold down the reveal and get the first stroke (I have the Stroke Order Animation Speed setting turned almost all the way down). After that, usually I remember it, can write it and move on. I also like that Skritter automatically grades it a stinky yellow 2, which fits right in with my normal grading scale.

This way, I at least get to write the majority of the character from memory, it was graded fairly, and it's fast and convenient with my Wacom.

Suggestion: Default the Stroke Order Animation Speed setting much, much slower

I'm really interested in knowing the answer to this question, too: which method is more effective. I've heard good arguments on both sides, which just means that it's that much more important to look at the data to find the truth.

Unfortunately, for Skritter there isn't currently an easy way that I can think of to do this segmentation. Ideally you'd be able to randomly assign items to one of the two groups and use the appropriate study mode depending on which group, but how would you do it?

Perhaps if you don't use hidden reading mode, you could quickly enough look at the pinyin to see if begins A-M or N-Z (just checked and that's a half-half distribution for pinyin in HSK words).

Later maybe we could design a better way to do this experiment. We've got more basic experiments to run first, though, including showing that Skritter is better than flashcards (which we're running this summer).

Thomas, tons of thanks for the idea with a 1 stroke reminder hint. Mind is really a strange thing - I got stuck for one quite a hard symbol, and then just by getting a hint for a first stroke I automatically wrote all symbol immediately.
That works!